Artificial Intelligence

2 Cornell U Teams Land up to $15 Million to Study AI, Autonomous Systems

• By Joshua Bolkan
• 07/24/17

Two research groups at Cornell University were recently awarded grants from the Department of Defense (DOD) of up to $7.5 million each to study artificial intelligence and autonomous systems.

Awarded under the DoD's Multidisciplinary University Research initiative (MURI), the grants will go to groups led by Robert Bruce van Dover, a professor in the Department of Materials Science and Engineering, and Hadas Kress-Gazit, an associate professor in the Sibley School of Mechanical and Aerospace Engineering. The grants will provide $1.5 million in funding each year for three years, with another two years of funding dependent upon progress and availability.

Van Dover and his team are working "to develop a multi-agent system that accelerates the science of materials discovery and development by integrating quantum physics principles, experimental materials synthesis, processing, characterization and AI-based algorithms," according to information released by Cornell.

The project, dubbed "Scientific Autonomous Reasoning Agent (SARA): Integrating Materials Theory, Experiment and Computation," aims to meet the goals of the Materials Genome Initiative, a project launched in 2011 "to create a new era of policy, resources and infrastructure that support United States institutions in the effort to discover, manufacture and deploy advanced materials twice as fast, at a fraction of the cost," according to the initiative's website.

Kress-Gazit and her team will work on "PERISCOPE: Perceptual Representations for Actions, Composition and Verification," a project designed "to develop autonomous systems capable of completing complex missions in unstructured and changing environments through the synthesis of ideas from control, perception, learning and verification," according to a Cornell news release.

The group will seek to create robots that are able to assess their own performance and make repairs. Instead of making assumptions about the environment that can't be modified, the group's robots will be able to verify or falsify assumptions and change their behavior accordingly.